Return to Projects
Our work focuses on time-sensitive AI, for example, to rapidly evaluate a photo for interesting content, or even to perform pipelines of intelligent tasks. Cascade is a Linux-based server (like a cloud function or lambda service) compatible with most existing AI dev tools (PyTorch, Tensor Flow, Spark, MXNet, etc) that can assist an AI by leveraging ultra-fast data paths transparently. With this approach, delay drops dramatically, throughput rises, and these properties drive the cost of operating a solution down.